Quote from: netmebtc on June 30, 2019, 09:45:39 AM

Quote from: Mashy81 on June 30, 2019, 02:01:53 AM

Quote from: netmebtc on June 30, 2019, 01:20:32 AM

Hello everyone, I have 6 vega64 (Samsung memory) rig windows10 18.5.1drive, core 1408/900mv, mem 1100/900mv, when I use the 0.5.2 version to mine XMR, each card can reach 2.4Kh, the wall power consumption of the whole machine is 1800 watts. Except for the power consumption of the motherboard of 200 watts, the average power consumption of a vega64 reaches 267w. Is this power consumption normal?

Is there a way to reduce the power consumption of vega to less than 185w, even if it is a little less than 2.4kh?

That power use is very high.
Undervolt. Power table mod. All the usualy things people do. Read a bit more of the thread and you will get plenty of ideas

I do some undervolt and ppt moding . set cclock to 1100/860mv mclock 1100/860mv , I got the 1450w power consumption at wall and 11.7Kh mining cnR

running trm 0.5.2 with cn_config 15*14:AAA

[2019-06-30 17:39:10] Mining cnr with 6 GPU workers
[2019-06-30 17:39:10] GPU 0 (06:00.0) 64 CUs Temp 63C Fan 49.00% (2408 RPM) Core 1062 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10] GPU 1 (19:00.0) 64 CUs Temp 63C Fan 49.00% (2446 RPM) Core 1058 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10] GPU 2 (13:00.0) 64 CUs Temp 51C Fan 58.00% (2874 RPM) Core 1055 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10] GPU 3 (16:00.0) 64 CUs Temp 63C Fan 52.00% (2575 RPM) Core 1053 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10] GPU 4 (03:00.0) 64 CUs Temp 62C Fan 56.00% (2788 RPM) Core 1069 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10] GPU 5 (09:00.0) 64 CUs Temp 62C Fan 53.00% (2599 RPM) Core 1040 MHz Mem 1100 MHz, VDDC 862.0 mV
[2019-06-30 17:39:10]
[2019-06-30 17:39:10] Current CN configs:
[2019-06-30 17:39:10] GPU 0 15*14:AAA
[2019-06-30 17:39:10] GPU 1 15*14:AAA
[2019-06-30 17:39:10] GPU 2 15*14:AAA
[2019-06-30 17:39:10] GPU 3 15*14:AAA
[2019-06-30 17:39:10] GPU 4 15*14:AAA
[2019-06-30 17:39:10] GPU 5 15*14:AAA
[2019-06-30 17:39:10]
[2019-06-30 17:39:10] Stats Uptime: 0 days, 00:16:53
[2019-06-30 17:39:10] GPU 0 [63C, fan 49%] cnr: 1.951kh/s, avg 1.951kh/s, pool 2.388kh/s a:24 r:0 hw:0
[2019-06-30 17:39:10] GPU 1 [63C, fan 49%] cnr: 1.949kh/s, avg 1.950kh/s, pool 1.393kh/s a:14 r:0 hw:0
[2019-06-30 17:39:10] GPU 2 [51C, fan 58%] cnr: 1.937kh/s, avg 1.945kh/s, pool 2.290kh/s a:23 r:0 hw:0
[2019-06-30 17:39:10] GPU 3 [63C, fan 52%] cnr: 1.939kh/s, avg 1.942kh/s, pool 1.892kh/s a:19 r:0 hw:0
[2019-06-30 17:39:10] GPU 4 [62C, fan 56%] cnr: 1.957kh/s, avg 1.962kh/s, pool 2.191kh/s a:22 r:0 hw:0
[2019-06-30 17:39:10] GPU 5 [62C, fan 53%] cnr: 1.928kh/s, avg 1.927kh/s, pool 2.388kh/s a:24 r:0 hw:0
[2019-06-30 17:39:10] Total cnr: 11.66kh/s, avg 11.68kh/s, pool 12.54kh/s a:126 r:0 hw:0

I modified the PPT with the registry.

GPU_P0=852;800
GPU_P1=991;810
GPU_P2=1084;820
GPU_P3=1138;830
GPU_P4=1150;840
GPU_P5=1202;850
GPU_P6=1212;850
GPU_P7=1100;860
Mem_P0=167;800
Mem_P1=500;800
Mem_P2=800;810
Mem_P3=1100;860
Fan_Min=1588
Fan_Max=4900
Fan_Target=63
Fan_Acoustic=2400
Power_Temp=75
Power_Target=0

The above parameters I have been through a number of tests, found a relatively stable data, if the voltage is set below 860mv, some cards will randomly appear crash

How can I further optimize and reduce the power consumption to 1200w?

If you are running the core at such so low frequency you should be able to lower the vcore to 810-820, but only if you also lower your HBM to 1028MHz or less. The reason is that the SOC will need more voltage, ie 860mV at 1107MHz than at 1028MHz. And the SOC frequency always needs to be higher or the same as the HBM. These are my settings for 4 Vega 56 pulling 650w from the wall:

http://m3server.se/img/QuadVega56XMR.png

kibatronic

Posts

Merits